Features of the USA300-HOU-MR genome

The USA300-HOU-MR genome is 2,872,915 bp in length and was annotated to encode 2658 proteins, 59 tRNAs (plus one pseudogene), 5 ribosomal RNA operons and 21 non-coding RNAs (Table 1). A circular representation of the sequenced genome is shown in Figure 3. To identify DNA regions unique to USA300-HOU-MR, two-way BLASTN was performed against each of the other complete staphylococcal genomes (Figure 4). USA300-HOU-MR carries the staphylococcal cassette chromosome mec (SCCmec) type IVa methicillin resistance cassette. In addition to SCCmec IVa, the strain also contains a ca. 31 kb region of DNA from 57860 to 88845 that is unique only to USA300-HOU-MR, USA300-FPR3757 and to S. epidermidis ATCC12228 (Additional file 1) [25]. This region maps immediately downstream of SCCmec and the entire element (34173 to 88845) is flanked by an IS431mec transposase gene and a possible transposase gene. The region downstream of SCCmec carries genes encoding an arginine deiminase operon, a Crp family transcriptional regulator, an APC family arginine/ornithine antiporter, and an arginine repressor. In USA300-FPR3757, this region is called ACME, for arginine catabolic mobile element [24]. Downstream of the arginine element we also annotated genes encoding a universal stress protein, a lysophospholipase, a possible S-adenosyl-L-methionine (SAM)-dependent methyltransferase, an ABC superfamily ATP binding cassette transporter, a P-ATPase copper transporter, a possible lipoprotein and an integrase. The ABC transporter proteins had top BLAST hits to nickel, dipeptide or oligopeptide transporters in other species and it was annotated as an oligopeptide transporter by Diep, et al. [24]. ClustalW analyses of the individual components of these transporters indicated a clustering of nickel-specific sequences with those from USA300-HOU-MR (data not shown). Thus, we believe that this transporter is likely involved in nickel acquisition. InterProScan, PFP and PSORTB, SecretomeP2.0 were used to augment the annotation data obtained by BLASTP and CDD. Based on this analysis, the SCCmec-ACME region encodes four non-classically secreted proteins and three other extracellular proteins. The region also encodes a possible endonuclease, possible tRNA ligases, RNA polymerases and a primase. We also identified a number of ORFs that were not annotated in USA300-FPR3757; these were hypothetical or conserved hypothetical proteins.

Start	Stop	Definition
Riboswitches
1,556,144	1,556,024	FMN riboswitch
1,899,283	1,899,148	FMN riboswitch
1,653,467	1,653,370	GcvT riboswitch
2,272,812	2,272,596	GlmS riboswitch
1,430,494	1,430,669	lysine riboswitch
1,783,037	1,782,862	lysine riboswitch
436,036	436,135	purine riboswitch
893,057	893,160	SAM riboswitch
1,025,974	1,026,102	YybP-YkoY riboswitch
Small ncRNAs
1,736,694	1,736,363	6S RNA
508,051	508,326	bacterial signal recognition particle RNA
1,930,023	1,929,680	noncoding RNA SprA
1,946,423	1,945,949	noncoding RNA SprB
1,959,339	1,958,106	noncoding RNA SprC
1,958,812	1,958,861	noncoding RNA SprC
2,087,533	2,086,851	noncoding RNA SprD
2,088,540	2,089,062	noncoding RNA SprE
2,090,767	2,090,977	noncoding RNA SprF/SprG
1,516,390	1,515,994	ribonuclease P (RNase P)
2,147,378	2,147,086	RNAIII
868,423	868,737	tmRNA SsrA

Pathogenicity islands and prophages

USA300-HOU-MR carries two complete prophage sequences, φSLT-USA300 and φβC-USA300. Prophage remnants also occur within pathogenicity island SaPI5. The φSLT-USA300 sequence is 98% identical to the 46139 bp φSLT sequence [28], excepting gaps greater than 500 bp. These gaps represent regions in USA300 encoding a helicase and virulence-associated protein E, two DNA polymerases, and two prophage regulatory proteins. The prophage carries the Panton-Valentine leukocidin genes lukS-PV and lukF-PV. By PHYLIP analysis, the φSLT-USA300 sequence is most similar to the prophages in MW2 and clinical strain A980470 (phiSLT) [28] (Figure 5a). As observed for φSLT-A980470 and φ-PVL, the φSLT-USA300 attachment site lies within a conserved hypothetical protein gene [28, 29]. The second prophage, φβC-USA300, is very similar to β-hemolysin-converting prophages identified in JH1, JH9, MRSA252, MSSA476, Mu50, MW2, N315, Newman, and NCTC8325 (Figure 5b). It is most similar to those found in JH1, JH9 and N315. In USA300, this prophage encodes staphylokinase, chemotaxis inhibitory protein and staphylococcal complement inhibitor; the order of these genes places it in immune evasion cluster (IEC) type B [30]. Unlike COL, USA300-HOU-MR lacks the L54α-like bacteriophage. As a result, the glycerol ester hydrolase, or geh lipase gene, which is disrupted in strains carrying the L54α-like prophage, is intact and presumed to be expressed.

Plasmids

Two plasmids were detected in USA300-HOU-MR (Figure 6a and 6b). The 3.1 kb plasmid, pUSA01-HOU, is identical to plasmid pUSA01 in USA300-FPR3757. It encodes a replication protein plus several conserved hypothetical proteins. The second larger plasmid, pUSA300-HOU-MR, appears to be a mosaic of parts of three plasmids: pN315 from S. aureus N315; pSR1 from S. aureus strain 01A103; and pSERP from S. epidermidis RP62A. The plasmid carries two recombinase genes and three IS431 transposase genes that may have played a role in fusion of the three elements. The replication region appears to be derived from pNF315B. pUSA300-HOU-MRSA encodes resistance to beta-lactam, macrolide, aminoglycoside, streptothricin (nourseothricin) and bacitracin antibiotics, consistent with the resistance pattern observed for USA300-HOU-MR (Table 3), with the exception of nourseothricin. The streptothricin acetyltransferase and 3',5'-aminoglycoside transferase genes are adjacent and map between two IS431 transposase genes; long inverted repeats (ca. 700 bp) flank these genes. The repeats are most similar to the inverted repeats found in IS257 from S. epidermidis strain SK76 (Genbank SEU40386) though similar repeats are found on a number of plasmids from both S. aureus and S. epidermidis. Thus, it appears that this region constitutes a compound transposon that could also be part of a larger element carrying the bacitracin transporter genes. A ca. 400 nt repeat (which is a subset of the ca 700 bp repeat) also occurs within the IS431 transpose genes at ca. 12.5 and 21 kb on the map. pUSA300-HOU-MR also encodes a cadmium transporter CadD and regulator CadX, and at least three additional predicted transcriptional regulators including BlaI and BlaR.

Comparison of USA300-HOU-MR and USA300-FPR3757

An insertion of 159 bp in USA300-HOU-MR but not in USA300-FPR3757 occurs within a staphylococcal conserved hypothetical protein between 1857878 and 1858036. This insertion is a duplication of flanking sequence and maintains the reading frame of the protein. Another difference identified is a 47 bp region in USA300-FPR3757, within the repeated region of the IgG binding protein gene; again, these are absent from USA300-HOU. Additional short deletions in USA300-HOU-MR also occur within intergenic regions. Also of note is a frameshift in the gene encoding an AraC family transcriptional regulator in USA300 (742860–745009). This frameshift is not observed in any of the other sequenced S. aureus strains. In addition, we identified a frameshift in the cell wall anchor protein gene (2623054–2626400) not reported in the annotation of FP3757.

Comparison of the two USA300 sequences revealed 92 single nucleotide polymorphisms (SNPs) and two four bp. deletions (Additional file 2). We also discovered 53 sequencing errors in the FPR357 sequence. Fifty-five of the SNPs map within ORFs, while 18 map within intergenic regions (this is consistent with the fact that 80% of the genome is coding sequence) and 7 map within rRNA sequences. Twenty-one of the changes within reading frames are synonymous, 34 are non-synonymous and 7 caused frameshifts. Two of the SNPs may explain differences in antibiotic susceptibility. One, a T (FPR357) to C (HOU) change in the gyrA gene causes a leucine to serine change at amino acid 84. The serine-containing GyrA protein confers resistance to ciprofloxacin (FPR357), while leucine confers the wild-type, susceptible phenotype (HOU, Table 2) [32]. A second SNP presumed to be involved in fluoroquinolone resistance maps within the topoisomerase IV parC gene. Here, the USA300-HOU-MR amino acid 80 is the wild-type serine, while USA300-FPR3757 has a tyrosine that confers resistance [33]. Three non-synonymous SNPs occur within the IgG binding protein A. SNPs in intergenic regions could affect gene regulation: a four base deletion at 2,627,226 would place a SarA binding motif closer to the sarU promoter and could affect its transcription (Figure 7).

We compared the USA300 SNPs to all other sequenced S. aureus strains (Additional file 2). For 39 of the SNPs, the other strains had the USA300-HOU-MR allele, while in 34 the allele was identical to FPR3757. For the remaining SNPs, the other strains showed both alleles, with some being triallelic.

Comparison of USA300-HOU-MR and of USA300-HOU-MS

We generated a draft sequence of the methicillin susceptible strain, USA300-HOU-MS, using the 454 Life Sciences pyrosequencing technology [34]. This strain is susceptible to all the antibiotics tested except penicillin (Table 1). The sequences were assembled into 326 contigs with a total contig length of 2.8 Mb. We observed 32 SNPs between the HOU-MRSA and MSSA strains (Additional file 1). We also identified two regions of DNA that are found in USA300-HOU-MR but not in USA300-HOU-MS. In contrast, we observed only one region in the MS strain not present in the USA300-HOU-MR strain. This 733 bp contig had a short DNA match (44 bp, 100% identity) to several staphylococcal sequences and a partial protein match (50% length, 38% identity) to a transport accessory protein. The sequences missing from USA300-HOU-MS included the SCCmec and ACME regions (34173–88845, Additional file 2). A second region not found in USA300-HOU-MS (680132 to 694219) includes a large number of staphylococcal conserved hypothetical proteins (Additional file 2). Four of the proteins encoded by this region may be secreted (two non-classical). Also encoded are a possible lipoprotein, a peptidoglycan hydrolase, a cell division protein, ATP- and GTP-binding proteins, a replication protein, a possible secreted actin-binding protein, a possible ornithine carbamoyltransferase and a possible histidinol dehydrogenase.

We surveyed other sequenced staphylococcal genomes using TBLASTX for the presence of ORFs that were homologous to proteins encoded by the regions of USA300-HOU-MR that were missing in the MSSA strain. These are shown in Additional file 2. With the exception of a few matches in the MW2 genome, the region from 51299 to 62735 seems to encode proteins unique to USA300-MR. This region also encodes four hypothetical and six conserved hypothetical proteins; several of these, although present, were not annotated in USA300-FPR3757. As previously discussed, this region may encode several secreted proteins. Though the arginine deiminase DNA region is unique to USA300 and to S. epidermidis ATCC 12228, genes encoding homologues of the proteins are found elsewhere in the USA300-HOU-MS genome and in most of the complete staphylococcal genomes. Proteins weakly similar (e value > e-10) to the nickel transporter were found in all staphylococcal genomes but this is likely due to generalized matches to other ABC transporter subunits. In contrast, the nickel transporter proteins are most homologous to proteins encoded by Staphylococcus haemolyticus. The proteins encoded by the region between 680132 and 694219 are also encoded by the genomes of two nosocomial MRSA strains, COL and MRSA252. These were not found in the CA isolate MW2.

USA300-HOU-MS does not carry plasmid pUSA300-HOU-MR but it does carry a 21 kb plasmid (pUSA300-HOU-MS) that encodes resistance to beta-lactams and to cadmium (Figure 6c). It also carries genes that are predicted to encode a bacteriocin and its associated immunity protein. The region from 9382 to 13575 on pUSA300-HOU-MS is nearly identical (95%) to nts 7644 to 11837 of plasmid pN315B. This corresponds to the region encoding the beta-lactamase gene and its regulators, BlaR and BlaI, and a Tn552 invertase and recombinase. In addition, this same region on pUSA300-HOU-MS is 96% identical to nts. 7615–13573 on pUS300-HOU-MR. Thus, pUSA300-HOU-MS also appears to be mosaic. We also identified sequences homologous to pUSA01 on two contigs that were 1212 and 1278 bp in length, respectively. Since pUSA01 has now been sequenced twice, we did not attempt to complete the pUSA01 sequence in the MS strain.

Comparison with other completely sequenced staphylococcal genomes

We used Mauve [35] to align the sequences of the seventeen completed sequenced staphylococcal genomes (Figure 8) The alignment revealed that USA300 clusters with COL, NCTC8325, and Newman strains and that there are at least four clades of MRSA circulating. This clustering suggests that not all methicillin resistant S. aureus strains are intimately related further suggesting that different clones have given rise to groups of methicillin strains.

USA300 isolates

Clinical isolate TCH1516 (USA300-HOU-MR) was obtained from an adolescent patient with severe sepsis syndrome [15]. The strain was representative of the predominant Sma I pulsed field gel electrophoresis (PFGE) pattern (Figure 2) observed in our clonal studies on CA-MRSA at TCH [9] and is indistinguishable from the CDC type strain USA300.0114 obtained from the Network on Antimicrobial Resistance in Staphylococcus aureus [40]. The strain is sequence type 8 (ST8). Clinical isolate TCH959 (USA300-HOU-MS) was obtained from a 12 year old white female with a buttock abscess. The PFGE profile of the MSSA strain differed from USA300.0114 by two fragments. USA300-HOU-MS lacked a fragment of ca. 230 kb, which is present in the USA300-HOU-MR and the MS pattern included a 180 kb fragment absent in MR. Additionally, the ca. 300 kb fragment was slightly larger in the MS strain. FPR3757 was kindly provided by Francois Perdreau-Remington (University of California, San Francisco).

Antimicrobial susceptibilities

Antimicrobial susceptibilities were obtained by microbroth dilution assays using Clinical Laboratory Standards Institute (CLSI) guidelines [41] (Table 3). Unlike the other sequenced USA300 isolate [24], USA300-HOU is susceptible to clindamycin, tetracycline, ciprofloxacin and mupirocin.

DNA sequencing and genome assembly

Genomic DNA was purified from CsCl gradients [42] and DNA sequencing was performed by Sanger dideoxy whole genome shotgun (WGS) and by the 454 Life Sciences pyrosequencing strategies [34]. Genomic and plasmid DNA from USA300-HOU-MR was sheared to a size of 2 kb by nebulization, and cloned into a derivative of pUC18 [43]. The clones were used for WGS DNA sequencing to 8× coverage by using dye terminator chemistry, data were collected on ABI 3730 sequencers, and reads were assembled using the ATLAS assembler [44]. This sequence was supplemented with contigs assembled from shorter reads generated on a 454 Life Sciences GS20 sequencer to 15× coverage. This combined assembly was refined by sequencing PCR products to fill in gaps and to resolve ambiguous regions. DNA from USA300-HOU-MS was sequenced by the 454 method to 17× coverage and the sequences assembled using the 454 Newbler assembler. The complete sequences have been deposited at DDBJ/EMBL/GenBank under the following the accession numbers: CP000730 (USA300-HOU-MR); CP000731 (pUSA300-HOU-MR); CP000732 (pUSA300-HOU-MS). The USA300-HOU-MS has been deposited under the project accession AASB00000000. The version described in this manuscript is the first version, AASB01000000.

Glimmer [45] and GeneMark [46] were used independently to predict open reading frames (ORFs). Predicted proteins were compared to the nr database using BLASTP. Visualization of gene predictions was performed using the Genboree system [47] and the CONAN database [48]. Two individuals independently annotated each ORF and annotations were reconciled and converted to single entries within the database. Proteins lacking a homolog in the nr database were called hypothetical proteins, those with a homolog of no known function were called conserved hypothetical proteins and of these, if homologs were only in other staphylococci they were called staphylococcal conserved hypothetical proteins.

Sequence analysis

DNA comparisons were performed with BLASTN and BLASTZ. Protein sequences were analyzed by BLASTP and BLASTX and by the use of the predictive tools InterProScan [49], PFP [50], PSORTb [51], Helix-Turn-Helix Predictor [52], MEROPs [53], and the Tcdb Transport Classification database [54]. Non-classically secreted proteins (those lacking signal peptides) were predicted using SecretomeP2.0 [39]. Riboswitches and non-coding small RNAs were identified using BLASTN, the Riboswitch finder [55] and Riboswitch Explorer [56]. ClustalW [57] and Mauve [35] were used for multiple sequence alignments and PHLYIP bootstrapping and tree generation (evolution.genetics.washington.edu/phylip.html). Single nucleotide polymorphisms (SNPs) between USA300-MRSA-HOU and FPR3757 were verified by resequencing.